A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 10, 19:00 UTC
Remove unused `use_marlin` variable in `Mxfp4MoEMethod` (#31549)
Signed-off-by: vaibhav sourirajan <vs2787@columbia.edu>
Vaibhav Sourirajan committed
Jan 2, 2026 at 00:13 UTC
1f43c121d57a1fdb8915761450769b8b74dc4e56
1 file changed
-1
vllm/model_executor/layers/quantization/mxfp4.py
-1
@@ -240,7 +240,6 @@ class Mxfp4MoEMethod(FusedMoEMethodBase):
240
self.mxfp4_backend = get_mxfp4_backend(moe.is_lora_enabled)
241
242
self.marlin_input_dtype = None
243
- self.use_marlin = self.mxfp4_backend == Mxfp4Backend.MARLIN
243
self.max_capture_size = (
244
get_current_vllm_config().compilation_config.max_cudagraph_capture_size
245
)